Fracture of Synthetic Diamond M
Total Page:16
File Type:pdf, Size:1020Kb
Fracture of synthetic diamond M. D. Droty Ctystallume, 3506 Bassett Street, Santa Clara, California 95054 R. H. Dauskardt,” A. Kant, and R. 0. Ritchie Center for Advanced Materials, Materials Sciences Division, Lawrence Berkeley Laboratory, and Department of Materials Science and Mineral Engineering, University of California, Berkeley, California 94720 (Received 10 March 1995; accepted for publication 11 May 1995) The fracture behavior of synthetic diamond has been investigated using indentation methods and by the tensile testing of pre-notched fracture-mechanics type samples. Specifically, the fracture toughness of free-standing diamond plates, grown by chemically-vapor deposited (CVD) methods, was measured using Vickers indentations and by the use of disk-shaped compact-tension specimens; the latter method provides an evaluation of the through-thickness fracture properties, whereas the indentation method was performed on the nucleation surface of the sample. Measured fracture toughness (K,) values were found to be approximately 5-6 MPaJm by both methods, indicating that the fracture resistance of CVD diamond does not vary appreciably with grain size (within the certainty of the testing procedures). Complications, however, arose with the fracture-mechanics testing regarding crack initiation from a relatively blunt notch; further work is needed to develop pre-cracking methods to permit more reliable fracture toughness testing of diamond. 0 1995 American Institute of Physics. I. INTRODUCTION ited studies on the fracture toughness of polycrystalline diamond,29 primarily due to the difficulties in toughness The unique combination of physical and mechanical measurements, such that the relationships between micro- properties of synthetic polycrystalline diamond make it a structure and mechanical properties are not understood. promising material for many structural applications; these The fracture toughness, K,, provides the most realistic include the development of ultra-hard coatings (e.g., for hard assessment of the fracture resistance of a brittle material in disks, bearings or cutting tools), in bioprosthetic devices, and terms of a measure of the critical stress intensity (i.e., the even in the design of monolithic or composite engineering intensity of the ,local linear-elastic stress and deformation materials. For such applications, the superior properties of fields) to cause unstable (Le., catastrophic) fracture from a diamond include the highest values of hardness, stiffness pre-existing crack (see, for example, Ref. 4j. Previous frac- [Young’s modulus), and room temperature thermal conduc- ture toughness measurements on synthetic diamond have fo- tivity shown by any material, coupled with a low coefficient cused on two “approximate” procedures, specifically involv- of friction. However, for most practical uses of diamond, an ing indentation techniques,” and a “fracture mirror” additional engineering parameter of importance is the resis- method;3 a summary of results is listed in Table I. The in- tance to fracture, as characterized by the fracture toughness. dentation method involves an extension of the hardness test, As elaborated below, this is a difficult parameter to measure where the value of K, is determined from the length of the in brittle materials such as diamond owing to its extreme radial cracks which develop at the corners of the indentation values of hardness and stiffness. following penetration of the surface of the sample with a Despite complications in the fracture toughness evalua- sharp pyrimidical (Vickers) indenter under sufficient 1oad.j tion of diamond materials, it is important to characterize This method is easy to perform, but the measurement accu- their fracture behavior because of the variety of microstruc- racy is limited by uncertainties in the magnitude of the re- trues which may be produced by synthetic processing meth- sidual stresses generated in the vicinity of the intent, specifi- ods. For example, diamond grown by chemical-vapor depo- cally involving the material-dependent constant relating sition (CVD) produces columuar microstructures with hardness to toughness.6 Fracture mirror measurements, con- mixtures of grain orientations, typically with a very large versely, rely on identifying a critical flaw along the plane variation in grain size (Figs. 1 and 2); details associated with where fracture has occurred; this technique typically is diffi- the growth of diamond by CVD methods are discussed cult to perform for diamond materials as a result of their elsewhere.’ Each microstructural change, such as in grain complex microstructures (see, for example, Figs. 1 and 2). size, shape and orientation, grain-boundary reaction layers, Accordingly, the objective of this paper is to describe the presence of inclusions or porosity, may well yield a wide studies to measure the fracture toughness of free-standing range of fracture toughness values and (strength-limiting) CVD diamond (- 150-200 pm thick) plates using full-scale flaw populations. However, to date there have only been lim- fracture mechanics test methods with pre-notched compact- tension specimens;7-g results are compared with surface ‘IPresent address: Department of Materials Science and Engineering, Stan- measurements obtained using the approximate indentation ford University, Stanford, CA 94305. techniques, and previously reported’ indentation data. J. Appl. Phys. 78 (5), 1 September 1995 0021-8979/95/78(5)/3083/6~$6.00 0 1995 American institute of Physics 3083 Downloaded 26 Jul 2002 to 128.32.113.135. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp TABLE L Summary of fracture toughness K, measurements on diamond. Average Kc Method WlWm) Reference Comment Indentation 5.6 present work CVD nucleation surface Indentation 5.3 2 CVD growth surface Indentation 5 II single crystal Indentation 4 12 single crystal Compact-tension 6.3 present work CVD Fracture mirror 7 3 CVD RLG. 1. SEM micrograph of the growth surface ofa CVD diamond sample. confirmed by Raman spectroscopy measurements which did not indicate any significant shift in relative wave number II. EXPERIMENTAL PROCEDURES from the characteristic value.” A. Material Diamond plates were prepared for this study by growth B. Fracture-mechanics testing in a microwave plasma reactor on 50.8 mm diameter [loo] Disk-shaped compact-tension DC(T) specimens were polished silicon wafers. Wafers are prepared for deposition used for the toughness measurements; such samples are rou- by cleaning in solvents and scratching with a fine diamond tinely utilized for K, measurements7 in metals and interme- powder for nucleation enhancement. Diamond growth was tahics, and more recently in ceramics (see Refs. 8 and 9 for achieved at 2.45 GHz excitation in a mixture of - 1% meth- stress intensity and compliance solutions for this geometry). ane in hydrogen at a total flow rate of 200 standard cubic Specimens were prepared for mechanical testing by laser cut- centimeters per minute (sccmj. The total chamber pressure ting the center of a free-standing slab to a 25 mm diameter. was 30-70 Torr and applied (microwave) power was 500- Additional features were produced to accommodate me- 1500 W. Diamond growth proceeded until a thickness of chanical gripping, along with a slit centered between the - 150 to 200 pm was obtained. A free-standing CVD dia- gripping holes; the latter defines the plane for extension of mond plate was then derived from the sample by chemical the crack under load from the local concentration of stress at removal of the silicon wafer in acid etchants.?he plate con- the tip of the slot. A sharp notch was further defined at this tained high quality diamond as evidenced by the sharp char- location by laser cutting to half of the specimen depth, in acteristic peak of 1332 cm-’ in the Raman spectrum with a order to promote notch acuity and crack stability, which in- minimum of non-diamond carbon indicated by the absence creases with initial crack length. Crack lengths in these of a broad peak at -1500 cm-’ [Fig. 3). Ln addition, the specimens were continuously monitored in situ by electrical- well-faceted microstructure apparent from the growth surface potential measurements across a thin metallized (NiCr) is characteristic of CVD diamond slabs of a similar thickness gauge (- 1000 A thick) sputtered on the substrate side of the (Fig. 1). The through-thickness columnar grain morphology disk; this surface had a specular surface finish of better than (Fig. 2) revealed grain sizes varying from +l,um on the 0.25 ,um roughness. With this electrical-potential technique, nucleation side to -20 pm on the growth side. (surface) crack lengths can be determined to a resolution of Because of the lack of curvature of the free-standing films, residual stresses were reasoned to be minimal; this was 3 ,,,,,,‘,,,,,‘,,,, I ,,‘,’ I 0 ~~‘~~~~‘.~~~‘~.~~‘~I.~ 1300 1400 1500 1600 1700 Relative Wavenumber [cm-‘] FIG. 2. SEM micrograph of a cross-section of a CVD diamond sample, showing the cohmmar microstructure with a very wide range of gram sizes FIG. 3. Raman spectrum of a CVD sample, showing a fully diamond (sp3 from 41 pm on the nucleation surface to -20 pm on the growth surface. bonding) structure. 3084 J. Appl. Phys., Vol. 78,.No. 5, 1 September 1995 Drory et a/. Downloaded 26 Jul 2002 to 128.32.113.135. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp Indentation load, P FIG. 4. Low magnification photograph of a diamond disk-shaped